2

Resources for Information Technology Research

The resources needed for research include funding and human capital, which are interrelated. Increases in funding for information technology (IT) research can enable industrial and university laboratories to hire more researchers, increase the number of graduate students trained in the nation's research universities, and allow the purchase of more IT hardware, software, and services to support those people. Similarly, increasing the size of the research workforce demands additional financial resources for salaries and technical infrastructure. But numbers alone do not tell the whole story. Equally important are the types of work supported and the types of organizations that fund or undertake the research. Vendors of computing and communications products, systems integrators, and end users all have different perspectives on the IT challenges that need to be addressed, and these perspectives combine with those of government funders of research and the researchers themselves to influence the scale and scope of the research agenda.

This chapter reviews trends in the nation's overall investment in IT research. The first section provides a framework for evaluating trends by explaining the importance of diversity in research portfolios, a theme carried forward through the chapter. The next two sections examine levels of government and industry funding for IT research, concentrating on the years 1987 to 1998. Distinctions are made, when possible, among funding sources (e.g., specific federal agencies, vendors, or users of IT components) and the types of research supported (e.g., component



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS 2 Resources for Information Technology Research The resources needed for research include funding and human capital, which are interrelated. Increases in funding for information technology (IT) research can enable industrial and university laboratories to hire more researchers, increase the number of graduate students trained in the nation's research universities, and allow the purchase of more IT hardware, software, and services to support those people. Similarly, increasing the size of the research workforce demands additional financial resources for salaries and technical infrastructure. But numbers alone do not tell the whole story. Equally important are the types of work supported and the types of organizations that fund or undertake the research. Vendors of computing and communications products, systems integrators, and end users all have different perspectives on the IT challenges that need to be addressed, and these perspectives combine with those of government funders of research and the researchers themselves to influence the scale and scope of the research agenda. This chapter reviews trends in the nation's overall investment in IT research. The first section provides a framework for evaluating trends by explaining the importance of diversity in research portfolios, a theme carried forward through the chapter. The next two sections examine levels of government and industry funding for IT research, concentrating on the years 1987 to 1998. Distinctions are made, when possible, among funding sources (e.g., specific federal agencies, vendors, or users of IT components) and the types of research supported (e.g., component

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS advances vs. system integration issues). The last section of the chapter reviews trends in academic IT research, which receives much of the government and industrial support. A credible discussion of research resources presupposes the existence of data: unfortunately, the present discussion is limited by the nature and quality of available statistics on IT research expenditures, as well as by lags between the time conditions are measured and the time they are reported. Despite extensive efforts by the National Science Foundation (NSF) and the Bureau of the Census, federal statistics on industrial and federally funded research remain difficult to track over time because some individual firms have been reclassified into different industry sectors and survey methodologies have been revised. Private sources of information, whether corporate reports or statistics from industry associations, typically do not distinguish expenditures on basic research from those on technology development (or applied research). Further complicating matters, neither federal nor private statistics speak to IT as a whole. Rather, they refer to academic disciplines (e.g., computer science and electrical engineering) or industry classifications (e.g., office and computing equipment and computing and data processing services). Some of the categories are being updated, but it is too soon to assess the impact of those changes. As computing and communications technologies converge and IT is infused into a growing number of products and services, assessing the size and needs of IT research will become even more difficult. Because of the limitations in the available data sets, this chapter does not attempt a definitive assessment. Instead, it presents and analyzes a mosaic of available statistics to elucidate the dominant themes in support for research. In some cases, funding for combined expenditures on research and development (R&D) is used as a proxy for research; in others, the distinctions in federal data between basic and applied research are used to gain some insight, however limited, into the overall investments in these areas. DIVERSITY IN THE RESEARCH BASE The payoff of any research, especially fundamental research, is inherently uncertain. Research managers cannot predict which projects will prove successful or produce the greatest benefit to their organizations, industry, or society as a whole. Accordingly, savvy research managers seek to invest in a range of diverse research programs as a strategy for ensuring that at least a fraction of the overall portfolio will pay off—preferably enough to justify the entire investment. The concept of preparing for the unpredictable by investing in diverse activities that pursue a

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS spread of the best possible ideas is known as “portfolio management” in financial markets, which rely on this concept to manage risk. Diversity within a research or financial portfolio plays much the same role as does genetic diversity in a species. A living organism carries a small amount of genetic material in addition to the genes that are essential for function. In a static environment, this genetic diversity imposes a cost beyond that carried by a species whose members are genetically identical and specifically tuned to exploit the environment. However, in times of a change or competition from other species, genetic diversity enables a species to adapt to the new environment using its extra resources. Similarly, diversity in the research base ensures that a nation's innovations will continue in the face of unforeseen changes in the technical, business, or societal landscape. 1 Diverse approaches will thus be available when changing conditions require new solutions quickly. No one can fully predict future needs. The need for high-speed packet switching could not have been predicted 25 years ago, yet it is the heart of the Internet today. The need for ultralow-power microprocessors was largely unexplored 20 years ago, whereas today it is a critical underpinning in the growing area of portable and handheld computing. The economic payoffs of specific investments are likewise difficult to predict. Coding theory and digital signal processing were important research areas 40 years ago because of their applications in telephony and military radio. There was no way to know, however, that this research would have such enormous importance for consumer cellular telephones and Internet multimedia conferencing, which have hundreds of millions of users. Lack of diversity in the IT research base can result from several factors. First, inbreeding can dilute the effectiveness of a research area as the same small community keeps funding and peer reviewing its members' projects. Or a research area can become too focused on a single approach that in retrospect turns out to have been unproductive. This can happen in a vigorous industry when firms adopt common approaches in their products, implying to researchers that even a successful new idea could not be introduced into practice. For example, radical new ideas for microprocessor designs might seem futile, because a tiny number of designs dominate today's market and the cost of market entry is enormous. However, low-power designs, crafted for the battery-powered portable devices that are sure to increase in number, might offer an opportunity for a very different approach. Second, there can be a lack of funding for certain types of research —resulting in an absence of understanding of some technology that might suddenly become important to the field. This can happen when innovation moves ahead of research and new products and services are developed without sufficient intellectual underpinnings. Arguably, some of

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS the problems inherent in large-scale systems fall into this category (as discussed further in Chapter 4). Third, research can reflect, to too great a degree, the objectives of the funders rather than the ideas of researchers. Although narrowly defined project goals can sometimes drive research that has serendipitous outcomes, they typically undermine diversity. This prospect fueled debate within the IT research community in the 1990s, when the High Performance Computing and Communications Initiative focused attention on the nature of research program definition and its impacts. Of course, it can be difficult to quantify the opportunity cost, which, at best, may be measurable only in retrospect. FEDERAL SUPPORT FOR INFORMATION TECHNOLOGY RESEARCH The federal government has been a strong supporter of IT research since World War II. Some of this research is conducted in federal laboratories, such as those supported by the Department of Energy (DOE) and the Department of Defense (DOD), but the greatest impact may have come from federally funded research carried out in university and industrial laboratories (CSTB, 1999). Over the past 50 years, this research has contributed to a wide range of important developments, including interactive time-shared computing, computer graphics, artificial intelligence, relational databases, and internetworking (CSTB, 1995, 1999).2 These technologies laid the foundation for new firms and new industries that have made substantial contributions to the nation's economic and social development. The context for decisions about new IT research continues to change—the industrial context seems particularly uncertain as this report is written—but broad lessons can be extracted from history to inform future decision making. As noted in earlier reports by the Computer Science and Telecommunications Board (CSTB, 1995, 1999), federal support for IT research has been most effective when (1) directed toward fundamental research with long-term payoffs, (2) used to support experimental prototypes that pushed the technological frontier and created communities of researchers that crossed institutional boundaries, and (3) expanded on research pursued in industry laboratories. Such investments not only generated new technical ideas and knowledge that subsequently were incorporated into new products, processes, and services, but also—especially in the case of university research—trained generations of researchers who went on to lead the IT revolution. Continued federal support for projects that complement industry-funded research in these ways will help maintain the strength of the IT sector. Federal agencies also need to continue to

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS look forward, supporting computing and communications research in areas that are likely to grow in importance. The following sections examine trends in federal funding between 1990 and 1998, the relative contributions of particular agencies, the operating styles of major federal agencies, and the characteristics of some large IT research programs. That period was chosen because it constitutes the most recent period for which consistent statistics are available. Other sections draw on data for 1999 and 2000. Trends in Federal Funding Trends in federal support for IT research over the past decade can be gleaned from data on funds obligated for research in computer science and electrical engineering, the two academic disciplines most closely associated with IT. Computer science encompasses the study of the theory of computing; the design, development, and application of computing devices; information science and systems; programming languages; and systems analysis—all topics that are directly applicable to IT. Electrical engineering includes the study of electronic devices and communication systems, which is directly relevant to IT, as well as the study of electric power systems, which is not. The sum of research expenditures for computer science and electrical engineering is an imperfect, but reasonable, proxy for IT research. Although it overstates federal expenditures by including work on electric power, this overstatement is offset by uncounted research in other academic disciplines relevant to IT, such as mathematics and cognitive science (important for understanding human-computer interaction). The data indicate that federal funding for IT research has, in general, been strong over the past decade. Combined federal funding for computer science and electrical engineering grew from $1.4 billion to $2 billion in constant dollars between 1990 and 1998—a 40 percent increase in real terms (Figure 2.1).3 However, federal funding for IT research remained virtually unchanged in real terms between 1993 and 1997, when the Internet and the World Wide Web began to exert a significant influence on the nation's economic and social structure, and when combined sales of IT goods and services were growing at an annual rate of more than 10 percent in real terms. 4 In other words, the explosion in IT applications throughout industry, government, and society was not matched by a commensurate increase in federal research support for the field—even as those applications began pushing beyond the knowledge limits of the underlying technology and began opening up new research opportunities. Despite the gains in funding for IT as a whole, federal support for research in electrical engineering appears to have declined between 1990

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS FIGURE 2.1 Federal funding for IT research, 1990 to 1998. SOURCE: National Science Foundation (2000a). and 1998. The data suggest a 16 percent drop in real terms, from $764 million to $639 million in constant dollars (Figure 2.2). Most of the cutswere in the applied research budget, effectively boosting the share of funds devoted to basic research from 23 percent to 34 percent of total research spending in electrical engineering. However, support for basic research did not grow appreciably until

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS FIGURE 2.2 Federal obligations for research in electrical engineering, 1990 to 1998. SOURCE: National Science Foundation (2000a). in electrical engineering appear to be due almost entirely to cutbacks in support from the DOD, which accounted for as much as 84 percent of the nation's total research funding for the field during this time period. This observation suggests that a significant portion of the cutbacks was directed toward those areas of electrical engineering that are directly related to IT, but sufficiently detailed data are not available to confirm

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS 1998. Reductions in total research this statement. Nor can available statistics reveal whether apparent reductions in funding for electrical engineering resulted from the reclassification of some research from electrical engineering to computer science. In computer science, combined federal expenditures for basic and applied research (also referred to as “total research”) more than doubled in real terms between 1990 and 1998, growing from $671 million to $1.4 billion in constant 1998 dollars (Figure 2.3). Here, however, funding for applied research grew more quickly than that for basic research. Although FIGURE 2.3 Federal obligations for research in computer science, 1990 to 1998. SOURCE: National Science Foundation (2000a).

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS spending on basic research increased from $269 million to $419 million during the 8-year period, its share of total federal funding for computer science research fell from 40 percent to 30 percent. Statistics on basic versus applied research must be used with caution because distinctions between the two categories are notoriously difficult to make and may only reflect differences in accounting methods. Nevertheless, the data correlate with the testimony to this committee (and others) by IT researchers—especially university researchers—who perceive a decided shift in federal funding away from fundamental research and toward more applied projects with narrower scopes of inquiry, additional project milestones, mandatory system demonstrations, and interim deliverables. This trend has an upside and a downside: it may enhance the accountability of government agencies and help document the benefits of public investments (objectives set forth by the Government Performance and Results Act), but IT researchers report that it hampers their ability to conduct long-term research with inherently uncertain outcomes. At risk is the type of work that has been the cornerstone of federally funded IT research for decades. Sources of Federal Support Despite growth in the number of agencies listed as supporting IT research, federal funding remains concentrated in a handful of agencies. As recently as 1998, 88 percent of all federal funds for computer science research were distributed by just three federal agencies: the Department of Defense (DOD), the Department of Energy (DOE), and the National Science Foundation (NSF). The DOD alone contributed 40 percent of the total, with much of its funding coming from the Defense Advanced Research Projects Agency (DARPA) (see Table 2.1). A similar proportion of all funding for basic research in computer science came from the same three agencies, with the NSF alone contributing 62 percent of the total in 1998. This funding pattern continues a historical trend: as far back as 1976 (the earliest date for which consistent data are available), these three agencies contributed 91 percent of federal funding for computer science research, with the DOD alone contributing 68 percent. Although the DOD has driven many important IT innovations in the past 50 years, the field's reliance on this one agency makes IT research support especially sensitive to fluctuations and directions in defense spending—and to repeated calls for research to be more relevant to defense missions. Defense budgets declined significantly in the post-Cold War environment, with total defense R&D declining 24 percent in real terms from its high in 1989 to 1999.5 Although DOD funding for computer science research grew 32 percent in real terms between 1990 and

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS TABLE 2.1 Federal Funding for Computer Science Research by Agency, 1998   Total Research Basic Research Agency Millions of Dollars Percent of Total Millions of Dollars Percent of Total Department of Defense 562 40 85 20 Department of Energy 396 28 22 5 National Science Foundation 267 19 258 62 Department of Health and Human Services 66 5 35 8 Department of Commerce 58 4 1 0 National Aeronautics and Space Administration 26 2 17 4 Other 25 2 1 0 Total 1,399 100 419 100 SOURCE: National Science Foundation (2000a). 1998, it varied from year to year, declining in real terms in 1991, 1994, and 1996. Steep increases in spending by NSF and DOE during those years more than compensated for fluctuating military funding, but increases in computer science spending were not matched in spending for electrical engineering. The DOD funding for electrical engineering research dropped 20 percent in real terms between 1990 and 1998, driving the decline in total federal funding for the field. The concentration of federal IT research funding within three organizations may have other limiting effects, not only on technology but also, perhaps, on the performance of government operations. Several agencies, including the Department of Health and Human Services (DHHS), the Social Security Administration, the Federal Aviation Administration, and the Internal Revenue Service, find their missions increasingly reliant on capable IT systems, which figure not only in internal processes (e.g., determinations and tracking of Social Security benefits) but also, increasingly, in the conduct of external activities for which they have some responsibility and in the very fabric of their relationships with external parties, from entities in regulated industries to individual citizens. The potential benefits to government agencies from IT are such that special federal efforts have been mounted to pursue them, including the Digital Government program described in Chapter 4. This level of attention and effort distinguishes IT from other types of infrastructure, such as transportation, on which agencies also depend.

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS Achieving the benefits of IT within agencies has been more difficult than articulating their promise. The difficulties these agencies have experienced with systems modernization over the past decade (see Chapter 3) suggest that adequate solutions to their IT needs are not available and cannot be developed within existing time and budget constraints. The problems are related mostly to the large scale of the systems and applications—a common issue that might have the best chance of being resolved if the agencies supported the relevant IT research. Yet the agencies are not mounting significant IT research programs, nor have they rushed to support the exploratory Digital Government program launched by the NSF to couple IT researchers to agencies with IT challenges—an initiative that could stretch the state of the art. Surely, support for IT research from federal agencies other than DOD, DOE, and NSF has increased as a percentage of total federal support since 1990, but the fraction remains small. Only 12 percent of total IT research funding and 13 percent of basic research support came from other agencies in 1998—and most of that was provided by science-based agencies such as DHHS and the National Aeronautics and Space Administration (NASA), which have long histories of attention to IT. Few other agencies have established IT research programs, and without the impetus of their resources and problem definitions, technical progress on these problems will probably be a matter of chance. Styles of Federal Support Federal agencies support IT research in different ways that tend to be suited to different types of problems. The most notable distinction is that between the research management styles of DARPA and the NSF. Research at DARPA has always emphasized the design and engineering aspects of IT, and building and experimenting with research prototypes are an essential aspect of that research.6 Such experimental work requires continuity—more funding per investigator and longer projects than are necessary for theoretical or paper studies (CSTB, 1994). DARPA's program managers assemble and oversee research portfolios within particular thematic areas. Researchers themselves play an indirect role in setting the objectives of the program through their interactions with program managers, and some DARPA programs specifically allow investigator-initiated proposals within the research theme of the program. Program managers are technically savvy—usually researchers on leave from university or industry—and know both the field and its researchers well. They work with research leaders and military leaders to develop long-term objectives that are both tractable for the research community and valuable to the DOD, balancing near-term military needs against longer-

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS academia or largely developed there. Significant examples include the Internet, reduced-instruction-set computing, redundant arrays of inexpensive disks for storage, object-oriented programming, CAD of integrated circuits, and computer graphics. Universities can be particularly important performers of fundamental and long-term research. Unlike industrial research, most university research is conducted in the public domain. Results of university research are disseminated widely throughout the research community, maximizing the impact of the research, and university graduates serve as key conduits of technology transfer as they move into jobs in other universities, government, and industry. Universities are fertile sources of innovation; free from pressures to make a near-term impact on the next generation of products, they often provide new ideas for established companies and seed the establishment and growth of numerous start-up companies. Maintaining the strength of university research is therefore key to ensuring the vitality of the IT industry. The following sections discuss trends in support for university research, gaps in such research, and commercialization of the research results. Trends in Support for University Research The available statistics present a mixed picture of funding for university research focusing on IT. Universities report that, between 1990 and 1998, constant-dollar funding for R&D in computer science grew from $614 million to $754 million, and constant-dollar funding for R& D in electrical engineering grew from $791 million to $1.02 billion. Approximately two-thirds of those funds came from federal sources, with the balance coming from industry, state and local governments, university funds, and other sources. Statistics on federal funding for university research indicate that federal support for IT-related research in universities grew at an average annual rate of 3.3 percent between 1990 and 1998 (Figure 2.5). But these statistics indicate that the rise is attributable almost entirely to increases in federal funding for computer science research, which expanded from $336 million to $470 million during the period of interest; federal funding for university research in electrical engineering rose at a rate of only 0.9 percent between 1990 and 1998 (from $165 million to $177 million) and actually declined in real terms after 1993.44 Additional IT-related research is conducted in university departments other than computer science and electrical engineering, but it tends not to be captured fully in federal statistics. Historically, this work has been pursued in science and engineering departments and has been directed toward large simulations of physical phenomena and technological systems. It has been a primary driver for research into high-performance

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS FIGURE 2.5 Federal funding for university research in IT, 1990 to 1998. SOURCE: National Science Foundation (2000a). computing and parallel processing. More recently, the number of departments engaged in IT-related work appears to have grown as IT has become more deeply ingrained in science and engineering, as well as a host of nontechnical fields. Business schools and departments of industrial engineering, for example, are studying the ways in which IT affects business processes. Medical schools and biology departments are conducting

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS research to enable better use of IT in providing patient care and in sequencing the human genome. The Department of Aeronautics and Astronautics at the Massachusetts Institute of Technology recently hired a computer scientist (Nancy Leveson) with expertise in software safety. As discussed in greater detail in Chapter 4, a number of universities have established new schools or departments to investigate issues at the intersection of IT and the social sciences. In all of these cases, it is difficult to determine the extent to which the work advances the state of the art in IT (i.e., should be considered IT research) versus the extent to which it is used to advance research in another discipline (i.e., supports development of IT systems to support research in another discipline). This report argues that there is great value in the former. Industry support for university research has grown over the past decade but still represents less than 10 percent of all university research funding. Moreover, it tends to be concentrated at a select set of universities. At Carnegie Mellon University, MIT, Stanford University, and the University of California at Berkeley, funding from industry constitutes 20 to 30 percent of IT funding for research. Such support can take several forms. Companies may sponsor research of potential interest to them, providing support for a faculty member and graduate students, or they may participate in collaborative programs in which industrial and academic researchers work side by side to bring new technology to market.45 Organizations such as the Semiconductor Research Corporation (SRC), whose members include most of the nation's largest manufacturers of integrated circuits, pool research funds and make grants to universities for nonproprietary research that will help a range of member companies. In August 1998, for example, the SRC announced that it would establish six national Focus Centers with a total of $60 million per year in new funding to pursue long-term research of interest to the semiconductor industry.46 The trend toward IT-related start-ups originating in universities (discussed below) also fosters a type of collaboration. These varied forms of collaboration have a number of benefits: they can compensate for fluctuations in federal research budgets, increase the relevance of academic research, and, at times, generate revenues from licensing. Industry also benefits because academic research allows it to access new technologies of particular interest, keep abreast of new developments, and, perhaps most importantly, identify promising young researchers. Gaps in Academic Research To some extent, research conducted in academic research laboratories is aligned with the research agendas of its sponsors. Because much research funding in IT comes from government and industry, both of

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS which appear to allocate most of their resources to component research, academic research has been slow to respond to emerging requirements for interdisciplinary research connected to the large-scale systems and IT applications that are responding to business and societal needs. This is not to say that academia has failed to develop highly innovative programs to educate students and conduct research on interdisciplinary topics but simply that there is substantial room for improvement. Just as industry research can become compartmentalized along product lines and industry sectors, academic research can track individual disciplines too closely. Faculty members tend to be rewarded on the basis of their contributions to a particular field, so setting off in new directions can have adverse consequences. Universities face difficult problems in conducting research on networks and large-scale systems: primarily they lack access to large operational systems—most of which are owned and operated by private firms—as well as tools for simulating the performance of such systems. This problem has persisted for decades (CSTB, 1994), and its consequences have worsened as interest grows in the social applications discussed in this report. As the framers of federal networking research programs have long known, only large networks populated by real users demonstrate the behaviors that need to be studied and understood. Even if academic researchers gain access to these systems, it is extremely difficult, if not impossible, to change their operation for experimental purposes, because users and their applications demand stability and availability. This problem was first noted when the research community's use of the Internet grew rapidly in the 1980s; the commercialization of the 1990s only exacerbated the problem (CSTB, 1994).47 The limited ability to simulate such systems is reflected in the poor understanding of their behavior. Commercialization of University Research University students, professors, and researchers often start new companies to commercialize the results of their research. Universities also license technology to industry, especially since the passage of the Bayh-Dole Act of 1980, which allows universities to license technologies emerging from federally funded research programs. The large number of new companies created to sell products based on university research, and the thousands of licenses that universities grant to firms, testify to the dramatic impact of university research on the private sector—and the effectiveness of the nation's innovation system in converting research results into new products and processes. Across all industries, the number of start-up companies emerging from university research is growing rapidly. A 1998 survey by the Asso-

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS ciation of University Technology Managers (AUTM) reported that, since 1980, more than 2,200 companies had been created to commercialize the results of research conducted in U.S. and Canadian universities, research hospitals, and other research institutions (AUTM, 1998). Almost half of those companies had been created since 1993. In 1997, 258 of the 333 startup companies in the survey came out of university research. In 1996, only 248 start-ups were reported by all the institutions combined. Although the number of start-up companies is increasing, the percentage of technologies licensed to start-up (as opposed to established) companies is decreasing. From 1977 to 1993, 50 percent of licenses were granted to start-up companies. Since 1993, only 29 percent of licenses were extended to start-up companies, and 61 percent were extended to existing companies. The implications of this trend are as yet unclear, and further study is needed; the trend could signify greater recognition within established companies of the value of university research, or it could suggest established companies' growing dependence on university research. The IT industry is home to a large number of firms that emerged from university research. Stanford University, for example, gave rise to a number of well-known Silicon Valley companies, including Sun Microsystems and Cisco Systems. MIT also gave rise to a number of firms, ranging from Open Market, Inc., an e-commerce firm, to RSA Data Security, which specializes in public key encryption, and more recently Akamai, which streamlines the downloading of content from popular Web sites. The AUTM survey reports that MIT contributed to the creation of 17 start-up companies in 1997, second only to the University of Washington, with 25. A report by BankBoston found that MIT graduates and faculty had been involved in founding 4,000 companies that employed 1.1 million people and had annual world sales of $232 billion in 1995; 57 percent of the employment resulted from firms in electronics and instruments (BankBoston, 1997). Carnegie Mellon University has licensed technologies to many small software and robotics companies, as well as LYCOS, one of the well-known players in the Internet search engine market. The characteristics of start-up companies that arise out of academia vary significantly among universities. For example, both Stanford and the University of California at Berkeley have provided many new technologies to Silicon Valley, but their approaches are quite different. Berkeley professors have tended to remain in academia. At Stanford, by contrast, “it's almost expected that a successful faculty member will at some point start a company” (Hamilton and Himelstein, 1997), although an individual may return to Stanford after the company is well launched. Berkeley's style is to “develop technology, convince existing companies to use the ideas, and then go back and develop more technology” (Hamilton and Himelstein, 1997). This pattern seems to be changing

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS rapidly: in any given year about 10 percent of Berkeley's electrical engineering and computer science faculty members are on leave starting a company. Each pattern illustrates one way in which faculty and students migrate between universities and start-ups. It is too early to tell whether the late-1990s trend of faculty across the country leaving academia to establish start-ups will persist, but the prospect is debated actively among academics. All other things being equal, the trend raises questions about the long-term capabilities of universities. CONCLUSION This review indicates that the recent growth in spending on IT research does not alleviate all concerns about the nation's research enterprise. Several underlying trends could ultimately limit the nation's innovative capacity and hinder its ability to deploy the kinds of IT systems that could best meet personal, business, and government needs. First, expenditures on research by companies that develop IT goods and services and by the federal government have not kept pace with the expanding array of IT. The disincentives to long-term, fundamental research have become more numerous, especially in the private sector, which seems more able to lure talent from universities than the other way around. Second, and perhaps most significantly, IT research investments continue to be directed at improving the performance of IT components, with limited attention to systems issues and application-driven needs. Neither industry nor academia has kept pace with the problems posed by the large-scale IT systems used in a range of social and business contexts—problems that require fundamental research. With the exception of IBM, most companies involved in developing IT systems for end-user organizations invest little in research. Academic researchers also have tended to ignore work on large-scale systems and social applications because they require interdisciplinary teams and very large budgets and because it is hard for them to obtain access to operational systems for experimental purposes. New mechanisms may be needed to direct resources to these growing problem areas. REFERENCES Arrow, Kenneth. 1962. “Economic Welfare and the Allocation of Resources for Invention,” in Richard Nelson, ed., The Rate and Direction of Innovative Activity. Princeton University Press, Princeton, N.J. Association of University Technology Managers (AUTM). 1998. FY 1998 AUTM Licensing Survey. AUTM, Norwalk, Conn. BankBoston. 1997. MIT: The Impact of Innovation. MIT Technology Licensing Office, Cambridge, Mass., March. Available online at <http://web.mit.edu/newsoffice/founders/>.

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS Barclay, Tom, Jim Gray, and Don Slutz. 1999. Microsoft TerraServer: A Spatial Data Warehouse. Technical Report MS-TR-99-29, Microsoft Corp., Redmond, Wash., June. Available online at <http://research.microsoft.com/~gray/Papers/MSR_TR_99_29_TerraServer.pdf >. Buderi, Robert. 1999. “Into the Big Blue Yonder,” Technology Review, July/August:46-53. Carey, John. 1999. “An Ivory Tower That Spins Pure Gold,” Business Week, April 19, pp. 167-168. Christensen, Clayton. 1997. The Innovator's Dilemma. Harvard Business School Press, Boston. Cohen, W., and D. Levinthal, 1990. “Absorptive Capacity: A New Perspective on Learning and Innovation. ” Administrative Science Quarterly 35:128-152. Computer Science and Telecommunications Board (CSTB), National Research Council. 1992. Keeping the U.S. Computer Industry Competitive: Systems Integration. National Academy Press, Washington, D.C. Computer Science and Telecommunications Board (CSTB), National Research Council. 1994. Academic Careers for Experimental Computer Scientists and Engineers. National Academy Press, Washington, D.C. Computer Science and Telecommunications Board (CSTB), National Research Council. 1995. Evolving the High Performance Computing and Communications Initiative to Support the Nation's Information Infrastructure. National Academy Press, Washington, D.C. Computer Science and Telecommunications Board (CSTB), National Research Council. 1999. Funding a Revolution: Government Support for Computing Research. National Academy Press, Washington, D.C. Computer Science and Telecommunications Board (CSTB), National Research Council. 2000. The Digital Dilemma: Intellectual Property in the Information Age. National Academy Press, Washington, D.C. Congressional Budget Office (CBO). 1999. “Current Investments in Innovation in the Information Technology Sector: Statistical Background,” April. Available online at <http://www.cbo.gov>. Defense Advanced Research Projects Agency (DARPA). 1999. “Information Technology Expeditions,” BAA 99-07. Available online at <http://www.darpa.mil/ito/Solicitations/CBD_9907.html>. Dertouzos, Michael L. 1999. “The Future of Computing,” Scientific American 281(2):52-55. Gompers, Paul A., and Josh Cohen. 1999. The Venture Capital Cycle. MIT Press, Cambridge, Mass. Hamilton, Joan, and Linda Himelstein. 1997. “A Wellspring Called Stanford,” Business Week, Silicon Valley Special Report (August 26). Available online at <http://www.businessweek.com/1997/34/b354112.htm>. Hardy, Quentin. 1999. “Motorola's New Research Efforts Look Far Afield,” Wall Street Journal (June 17):B6. Information Week. 1999. “Information Week 500: Industries At a Glance,” September 27. Available online at <http://www.informationweek.com/754/99iw500.htm. Lerner, Josh. 1999a. “Small Business Innovation and Public Policy,” in Are Small Firms Important? Zoltan Acs, ed. Kluwer Academic Publishing, New York. Lerner, Josh. 1999b. “Small Business, Innovation, and Public Policy in the Information Technology Industry,” paper prepared for the conference Understanding the Digital Economy: Data, Tools, and Research, Washington, D.C., May 25-26. Markoff, John. 1999. “Microsoft Brings in Top Talent to Pursue Old Goal: The Tablet,” New York Times (August 30):C1,C10. National Science and Technology Council (NSTC), IT2 Working Group. 1999a. Information Technology Research for the Twenty-First Century: A Bold Investment in America's Future. Implementation Plan. National Coordination Office for Computing, Information, and Communications , Arlington, Va., June.

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS National Science and Technology Council (NSTC), Committee on Technology, Subcommittee on Computing, Information, and Communications R&D. 1999b. Information Technology Frontiers for a New Millennium: Supplement to the President's FY 2000 Budget, National Coordination Office for Computing, Information, and Communications , Arlington, Va., April. National Science Foundation (NSF). 1999. “Information Technology Research: Program Solicitation, NSF 99-167, ” September 28, available online at <http://www.nsf.gov/pubs/1999/nsf99167/nsf99167.htm>. National Science Foundation (NSF), Division of Science Resources Studies. 2000a. Federal Funds for Research and Development: Fiscal Years 1998, 1999, and 2000. Arlington, Va., forthcoming. National Science Foundation (NSF), Division of Science Resources Studies. 2000b. Research and Development in Industry: 1998. Arlington, Va., forthcoming. Nelson, Richard. 1959. “The Simple Economics of Basic Research,” Journal of Political Economy 67(2):297-306. Organization for Economic Cooperation and Development (OECD). 1999. STI Scoreboard of Indicators. OECD, Paris, France. Peltz, Michael. 1996. “High Tech's Premier Venture Capitalist,” Institutional Investor 30 (June):89-98. President's Information Technology Advisory Committee (PITAC). 1999. Information Technology Research: Investing in Our Future. National Coordination office for Computing, Information, and Communications , Arlingon, Va., February. Semiconductor Industry Association (SIA). 2000. The Silicon Century. Semiconductor Industry Association, San Jose, Calif. Smith, Douglas K., and Robert C. Alexander. 1988. Fumbling the Future: How Xerox Invented, Then Ignored the Personal Computer. William Morrow and Company, New York. Streitfeld, David. 1999. “Capital and Ideas: Financiers of the Information Age Serve Up Cachet, Cash,” Washington Post (August 15):H1. Takashi, Dean. 1996. “Intel Shifts Its Focus to Original Research,” WSJ Interactive Edition (August 26):1. Available online at <http://interactive3.wsj.com/edition/articles/SB841015179745727000.htm >. von Hippel, Eric. 1988. The Sources of Innovation. Oxford University Press, New York. White House. 2000. “Information Technology Research and Development: Information Technology for the 21st Century.” Press release dated January 21, Office of the Press Secretary, Washington, D.C. NOTES 1. Diversity is not, of course, the only factor in research success. The quality of the research is also of paramount importance. Quality can be assured through mechanisms such as peer review. 2. The Internet, for example, traces its roots to the DOD's ARPANET, built in the late 1960s and 1970s. Early work in virtual reality was supported by the government, and continued government investments in the technology sustained the field even when early commercial interest waned. Many of the most important advances in artificial intelligence came from government-funded research. 3. All data on federal funding for IT research in this paragraph were derived from the National Science Foundation (2000a). 4. The growth rate cited includes sales in five industry sectors defined in the standard industrial classification (SIC) codes: office, computing, and accounting machines (SIC 357),

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS communications equipment (SIC 366), electronic components (SIC 367), communications services (SIC 48), and computer and data processing services (SIC 737). 5. According to preliminary estimates from the National Science Foundation, defense R&D spending will decrease even further in FY00. 6. Many important, lasting IT developments sprang from DARPA's experimental projects, such as the ARPANET (which laid the groundwork for the Internet) and the Very Large Scale Integrated Circuit program, which helped advanced reduced-instruction-set computing. 7. Research supported by the NSF has contributed significantly to the evolution of IT. An important capability, scientific visualization, grew out of NSF sponsorship of computing in the service of science. Visualization, which uses carefully designed images to allow scientists and engineers to glean insight from computer simulations of natural phenomena, is now widely used in scientific computing and advanced engineering applications such as jet engine design. 8. For more information on the NGI, see <www.ngi.gov>. 9. The university centers established as part of ASCI are the Center for Integrated Turbulence Simulation at Stanford University, the Computational Facility for Simulating the Dynamic Response of Materials at the California Institute of Technology, the Center for Astrophysical Thermonuclear Flashes at the University of Chicago, the Center for Simulation of Accidental Fires and Explosions at the University of Utah, and the Center for Simulation of Advanced Rockets at the University of Illinois at Urbana-Champaign. 10. For more information on Project Oxygen, see Dertouzos (1999). 11. Additional information on the University of California at Berkeley 's Endeavor project is available online at <http://endeavor.cs.berkeley.edu>. 12. Additional information on the University of Washington's Portolano/Workscape project is available online at <http://portolano.cs.washington.edu/>. 13. As of April 2000, DARPA planned to transform its expeditions program into a program that would explore “ubiquitous computing,” a term used to describe the incorporation of computing and communications capabilities into a range of everyday devices. 14. It should also be cautioned that it is notoriously difficult to separate research from development, especially given that fundamental research advances sometimes emanate from focusing on development projects. Most often research and development are lumped together in the statistics, and attempts to separate out the research should be viewed with some skepticism. 15. The 20 percent figure reported in the 1998 data is unusually high, suggesting some inconsistencies in the collection or reporting of the data. IT firms reported that 10 percent of their research dollars were allocated to basic research in 1997, which is more consistent with earlier reports and anecdotal reports from research managers. 16. The Census Bureau is in the process of shifting from the SIC to a new system, the North American Industry Classification System, which features significant changes such as the introduction of an Information Sector and is undergoing additional modification and revision. Additional information on the transition to the new industry classification system is available online at <http://www.census.gov/epcd/www/naics.html>. 17. Despite the difficulties in tracing the movements of firms among industry sectors, federal statistics are still the best source of data for tracking research in the IT industry. Corporate annual reports and other public documents cannot be used because individual companies do not report research investments in these documents, although most list combined research and development investments. 18. Indeed, there is reason to believe that much of the decline in reported research and development investments in the office and computing equipment industry between 1990 and 1991 resulted from the reclassification of large firms to other industries.

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS 19. Combined R&D investments for these firms totaled $32 billion in 1998. R&D investments for all firms contained in the NSF survey of industrial R&D in the office and computing equipment, communications equipment, electronic components, communications services, and computing/data processing services industries in 1998 totaled $45 billion. 20. Many large IT firms were criticized in the 1980s and early 1990s for failing to take advantage of technologies developed in their own labs. Xerox, for example, developed one of the earliest personal computers (the Alto) but never successfully marketed it. See Smith and Alexander (1988). 21. Robert Metcalfe, a founder of 3Com Inc., has said that the value of the network scales as the square of the number of users. This is now called Metcalfe's law. 22. This is not to say that there will be no effort to displace the prevailing technology, as the open-source software movement and the Linux-based initial public offerings demonstrate. 23. Of course, there can be benefits to the rapid adoption of new technologies, and lock-in as well, in that they allow other innovators to build on top of a commonly accepted platform. It is only when limitations in the platform itself become evident and impede further innovation that lock-in becomes problematic. 24. Such companies have forced many of the traditional computer manufacturers, such as IBM, to streamline their PC operations, sometimes establishing them as separate business lines with their own cost structures. 25. A notable proponent of this theory is Christensen (1997). 26. For Motorola, which has roughly $30 billion in sales, this ratio would imply about $300 million in research funding. 27. For example, work in natural language processing has long-term goals, but it already has contributed to the grammar checker in Microsoft Office. 28. These activities correspond to SIC codes 7371 and 7373. 29. Andersen Consulting employs about 50,000 workers, so the research group represents just 0.4 percent of its workforce. 30. The information on Andersen Consulting's research activities was obtained from Joseph Carter, Andersen Consulting, in a presentation to the study committee in Palo Alto, California, on February 10, 1998. 31. The data on Lockheed Martin's R&D expenditures were obtained from B. Clovis Landry, vice president of technology, Lockheed Martin Information & Services Sector, October 11, 1999. 32. Personal communication from Irving Wladawsky-Berger, vice president of technology and strategy for IBM, October 6, 1999. 33. Amazon.com reported $47 million in product development expenses in 1998, most of which were related to continual enhancement of the features, content, and functionality of the company's Web sites and transaction processing systems, as well as investments in systems and telecommunications infrastructure. Merrill Lynch reported in 1997 that it would spend $200 million to complete the development of a technology platform for its financial consultants by the third quarter of 1998. 34. Needless to say, most of Boeing's $1.9 billion R&D budget is allocated to non-IT activities. 35. These data are from Amazon.com's annual 10-K report to the SEC. 36. Merrill Lynch reported in 1997 that it was investing some $200 million in the development of a new platform for its financial consultants called the Trusted Global Advisor system. In keeping with new accounting standards, Merrill Lynch amortized $72 million in development costs for internal-use software in 1998. These amounts are amortized over the useful life of the developed software (generally 3 years).

OCR for page 48
MAKING IT BETTER: EXPANDING INFORMATION TECHNOLOGY RESEARCH TO MEET SOCIETY'S NEEDS 37. Most of these patents have been awarded since 1998, although the patent applications were submitted several years before the awards. 38. In the late 1990s, end-user organizations also began applying for—and receiving—patents covering methods of doing business. Considerable controversy has arisen around this subject. The Computer Science and Telecommunications Board is developing a prospectus for a study of this issue. For additional background on the patenting of business practices, see CSTB (2000), especially pp. 192-198. 39. Anoop Gupta, a Stanford University professor on leave at Microsoft at the time, characterized this distinction to the committee on February 10, 1998, as follows: “The difference between black and white magic is really in its symbolism and intent.” Symbolism and intent seem to determine the perceptions of whether something is research or not. Whether knowledge is created is often overlooked. From this “intent-based” perspective, the work of start-ups is not research, whereas from the perspective of producing knowledge, it certainly is. 40. These data are from a PricewaterhouseCoopers MoneyTree survey. 41. Preliminary statistics from PricewaterhouseCoopers indicated that for the second quarter of 1999, 63 percent of VC investments went to firms in the communications, software and information, and computers and peripherals industries. 42. Data from VentureOne Corporation, as reported in Streitfeld (1999). 43. The definitions of seed, start-up, and expansion financing used here are derived from OECD (1999). 44. The apparent disparities between the research funding numbers reported by universities and by federal agencies are due largely to differences in the ways the surveys are administered to collect these data. 45. The federal government has attempted to stimulate collaboration between industry and academia as a means of improving the competitiveness of U.S. companies and of better exploiting the results of federally sponsored research. For example, NSF established the ongoing Engineering Research Centers program in the 1980s to foster partnerships among government, industry, and universities in research and engineering. This program is more fully described in Chapter 4 of this report. 46. Two Focus Centers had been established as of May 2000. The first is led by researchers at the University of California at Berkeley; the second, by researchers at the Georgia Institute of Technology. Each involves researchers from a number of other universities. Additional information on the program is available in SIA (2000). 47. Congressional hearings that predated the 1995 commercialization of the NSFnet featured debates over “experimental” versus “production” networks.